Re: crypting filesystems

[Felipe Alfaro Solana <felipe.alfaro@gmail.com>]

On Apr 4, 2005 9:51 PM, Wiktor <victorjan@poczta.onet.pl> wrote:
> Hi,
> 
> I'm using the following method and it seems to be working fine
> (involving crypto-loop):
> 
> i have normal ext3 /boot partition, where i store kernel image & initrd.
> after lilo boots the kernel, initrd sets up /dev/loop0 to be
> crypto-loop/blowfish for /dev/hda1 (losetup /dev/loop0 /dev/hda1 -e
> blowfish). losetup asks for passphrase, and (if entered correctly),
> /dev/loop0 is mounted as root filesystem (it can be done also by simple
> mount call: mount /dev/hda1 /some-place -o rw,encryption=blowfish). for
> encrypting more filesystems with one passphrase, you can read it in
> shell script in non-echo-mode (if such exists, i'm not sure), and pass
> it to mount or losetup. crypto-loop makes possible to switch encryption
> type without modifying whole initrd.
> 
> Regarding your questions:
> 
>  > 1. In order to put in the passphrase just once a time at booting, I
> put the passphrase in a gpg-crypted file (cipher AES256 and 256Bit key
> size), which is decrypted at boot-time to /tmp (-> tmpfs) and
> immediately removed with shred, after activating the three partitions.
> Is it possible to see the cleartext password after this action in tmpfs?
> 
> Disk encryption usually protects from hardware-attacks (when hacker has
> physical access to the hardware). if you keep passphrase
> reversible-encrypted, attacker can read it and run brute-force attack
> using some huge-computing-capacity. is this what you want?
> 
>  > 2. Is it possible to gain the passphrase from the active encrypted
> partitions (because the passphrase is somewhere held in the RAM)?
> 
> Only when attacker has root privileges. But i'm not sure if it is
> possible to extract passphrase knowing both encrypted and not encrypted
> data. What i mean is that usually each filesystem begins with
> filesystem-specyfic-header, which is constant or similar to each other.
> so, if attacker has encrypted form of this header and can estimate
> unencryptes form, it can possibly gain the passphrase. (but therse are
> only my ideas, i don't know how the encryptino-algorithm works).

What´s kept in RAM is the AES key used to decrypt disk blocks.
However, the passphrase from which the AES key is derived (usually by
using a hash function) is not kept in memory.